Decentralized learning is emerging as a scalable and privacy-preserving alternative to centralized machine learning, particularly in distributed systems where data cannot be centrally shared among multiple nodes or clients. While Federated Learning is widely adopted in this context, Knowledge Distillation (KD) is emerging as a flexible and scalable alternative where model output is used to share knowledge among distributed clients. However, existing studies often overlook the efficiency and effectiveness of various knowledge transfer strategies in KD, especially in decentralized environments where data is non-IID. This study provides key insights by examining the impact of network topology and distillation strategies in KD-based decentralized learning approaches. Our evaluation spans several dissimilarity measures, including Cross-Entropy, Kullback-Leibler divergence, Triangular Divergence, Jensen-Shannon divergence, Structural Entropic Distance, and Multi-way SED, assessed under both pairwise and holistic distillation schemes. In the pairwise approach, distillation is performed by summing the client-wise dissimilarities between a client’s output and each neighbor’s prediction individually, while the holistic approach computes dissimilarity with respect to the average of the output predictions received from neighboring clients.We also analyze performance across client connectivity levels to explore the trade-off between convergence speed and model accuracy. The results indicate that the holistic distillation approach, which averages client predictions, outperforms the sum of pairwise distillation, especially when employing alternative measures like TD, SED, and JS. These measures offer improved performance over conventional metrics such as CE and KL divergence.

Decentralized Prognostics and Health Management, particularly Remaining Useful Life (RUL) prediction, is essential for industrial systems operating under heterogeneous and privacy-constrained conditions. Data isolation across clients limits traditional centralized modeling, motivating the development of collaborative yet privacy-preserving solutions. This work investigates decentralized Knowledge Distillation (KD) as a communication-efficient framework for time-series-based RUL estimation. Each client independently trains a local model on private data and iteratively refines it by distilling knowledge from connected peers, using previous iteration weights for initialization. Three KD strategies, namely pairwise, holistic, and weighted, are evaluated on the C-MAPSS turbofan engine dataset. Preliminary results indicate that pairwise distillation improves cross-client generalization (+15.3%), while the weighted approach enhances local stability (+5.1%). Overall, decentralized KD delivers competitive predictive performance with minimal communication, providing a scalable, privacy-preserving framework for predictive maintenance.

The ACM Lifelog Search Challenge (LSC) is the only long-running benchmark that evaluates lifelog retrieval systems through real-time, human-in-the-loop interaction. This paper presents a longitudinal analysis of technical and interaction design progress at LSC from 2022 to 2024, conducted under fixed datasets, tasks, and evaluation protocols. Unlike prior surveys of lifelogging research, our study leverages competitive, synchronous evaluations to examine how retrieval architectures, interface design choices, and interaction strategies directly influence task performance. Through a comparative analysis of known-item search, question answering, and ad-hoc retrieval tasks, we identify a clear paradigm shift from concept-based pipelines toward embedding-driven and LLM-supported retrieval systems. We show how these shifts affect retrieval effectiveness, response time, and user behaviour, as well as provide empirical insight into how retrieval and interface design choices manifest in performance during interactive search. Beyond documenting emerging techniques, we expose evaluation artefacts such as performance variability across system instances and user familiarity effects, raising important considerations for future interactive benchmarks.

Relevance feedback mechanisms have garnered significant attention in content-based image and video retrieval thanks to their effectiveness in refining search results to better meet user information needs. This paper provides a comprehensive comparative analysis of four techniques: Rocchio, PicHunter, Polyadic Query, and linear Support Vector Machines, representing diverse strategies encompassing query vector modification, relevance probability estimation, adaptive similarity metrics, and classifier learning. We conducted experiments within an interactive image retrieval system, with varying amounts of user feedback: full feedback, limited positive feedback, and mixed feedback. In particular, we introduce novel enhanced versions of PicHunter and Polyadic search incorporating negative feedback. Our findings highlight the benefits of integrating both positive and negative examples, demonstrating significant performance improvements. Overall, SVM and our improved PicHunter outperformed the other approaches for ad-hoc search, especially in cases in which the feedback process is iterated several times.

Recent advancements in deep learning have significantly enhanced content-based retrieval methods, notably through models like CLIP that map images and texts into a shared embedding space. However, these methods often struggle with domain-specific entities and long-tail concepts absent from their training data, particularly in identifying specific individuals. In this paper, we explore the task of identity-aware cross-modal retrieval, which aims to retrieve images of persons in specific contexts based on natural language queries. This task is critical in various scenarios, such as for searching and browsing personalized video collections or large audio-visual archives maintained by national broadcasters. We introduce a novel dataset, COCO Person FaceSwap (COCO-PFS), derived from the widely used COCO dataset and enriched with deepfake-generated faces from VGGFace2. This dataset addresses the lack of large-scale datasets needed for training and evaluating models for this task. Our experiments assess the performance of different CLIP variations repurposed for this task, including our architecture, Identity-aware CLIP (Id-CLIP), which achieves competitive retrieval performance through targeted fine-tuning. Our contributions lay the groundwork for more robust cross-modal retrieval systems capable of recognizing long-tail identities and contextual nuances. Data and code are available at https://github.com/mesnico/IdCLIP.

In this paper, we propose and analyze Vec2Doc, a novel training-free method to transform dense vectors into sparse integer vectors, facilitating the use of inverted indexes for information retrieval (IR). The exponential growth of deep learning and artificial intelligence has revolutionized scientific problem-solving in areas such as computer vision, natural language processing, and automatic content generation. These advances have also significantly impacted IR, with a better understanding of natural language and multimodal content analysis leading to more accurate information retrieval. Despite these developments, modern IR relies primarily on the similarity evaluation of dense vectors from the latent spaces of deep neural networks. This dependence introduces substantial challenges in performing similarity searches on large collections containing billions of vectors. Traditional IR methods, which employ inverted indexes and vector space models, are adept at handling sparse vectors but do not work well with dense ones. Vec2Doc attempts to fill this gap by converting dense vectors into a format compatible with conventional inverted index techniques. Our preliminary experimental evaluations show that Vec2Doc is a promising solution to overcome the scalability problems inherent in vector-based IR, offering an alternative method for efficient and accurate large-scale information retrieval.

Relevance feedback is a well-established approach to refine search results based on user input, but its comparative evaluation across different methods remains limited in practice. This demonstration paper introduces an interactive platform that supports and compares four relevance feedback methods-Rocchio, PicHunter, Polyadic Search, and SVM-based active learning-under consistent conditions. The primary goal is to enhance the understanding of how different relevance feedback methods affect retrieval performance from both a technical and user-centric perspective. The source code is available at https://github.com/francescascotti16/Demo-Relevance-Feedback, while the demonstration can be found at http://relevance-feedback.isti.cnr.it/.

The ACM Lifelog Search Challenge (LSC) is an annual interactive competition that evaluates interactive systems for searching and exploring lifelog data. This paper reviews the recent advances in interactive lifelog retrieval as demonstrated at the ACM LSC from 2022 to 2024. Through a detailed comparative analysis, we highlight key improvements across three main retrieval tasks: known-item search, question answering, and ad-hoc search. Our analysis identifies trends such as the widespread adoption of embedding-based retrieval methods (e.g., CLIP, BLIP), increased integration of large language models (LLMs) for conversational retrieval, and continued innovation in multimodal and collaborative search interfaces. We further discuss how specific retrieval techniques and user interface (UI) designs have impacted system performance, emphasizing the importance of balancing retrieval complexity with usability. Our findings indicate that embedding-driven approaches combined with LLMs show promise for lifelog retrieval systems. Likewise, improving UI design can enhance usability and efficiency. Additionally, we recommend reconsidering multi-instance system evaluations within the expert track to better manage variability in user familiarity.

The ACM Lifelog Search Challenge (LSC) is a venue that welcomes and compares systems that support the exploration of lifelog data, and in particular the retrieval of specific information, through an interactive competition format. This paper reviews the recent advances in interactive lifelog retrieval as demonstrated at the ACM LSC from 2022 to 2024. Through a detailed comparative analysis, we highlight key improvements across three main retrieval tasks: known-item search, question answering, and ad-hoc search. Our analysis identifies trends such as the widespread adoption of embedding-based retrieval methods (e.g., CLIP, BLIP), increased integration of large language models (LLMs) for conversational retrieval, and continued innovation in multimodal and collaborative search interfaces. We further discuss how specific retrieval techniques and user interface (UI) designs have impacted system performance, emphasizing the importance of balancing retrieval complexity with usability. Our findings indicate that embedding-driven approaches combined with LLMs show promise for lifelog retrieval systems. Likewise, improving UI design can enhance usability and efficiency. Additionally, we recommend reconsidering multi-instance system evaluations within the expert track to better manage variability in user familiarity and configuration effectiveness.

This paper conducts a thorough examination of the 12th Video Browser Showdown (VBS) competition, a well-established international benchmarking campaign for interactive video search systems. The annual VBS competition has witnessed a steep rise in the popularity of multimodal embedding-based approaches in interactive video retrieval. Most of the thirteen systems participating in VBS 2023 utilized a CLIP-based cross-modal search model, allowing the specification of free-form text queries to search visual content. This shared emphasis on joint embedding models contributed to balanced performance across various teams. However, the distinguishing factors of the top-performing teams included the adept combination of multiple models and search modes, along with the capabilities of interactive interfaces to facilitate and refine the search process. Our work provides an overview of the state-of-the-art approaches employed by the participating systems and conducts a thorough analysis of their search logs, which record user interactions and results of their queries for each task. Our comprehensive examination of the VBS competition offers assessments of the effectiveness of the retrieval models, browsing efficiency, and user query patterns. Additionally, it provides valuable insights into the evolving landscape of interactive video retrieval and its future challenges.

The way we create, consume and interact with multimedia content has changed significantly in recent years with the advent of affordable recording devices and easy sharing and access in the form of mobile phones. With the imminent wave of affordable devices that enable mixed reality experiences and the large variety of devices on the market, interaction with multimedia content is expected to continue to evolve rapidly. This will also drastically affect the entire area of multimedia information retrieval in eXtended Reality (XR), for instance by novel ways to express user needs in VR, result presentation that takes the specific capabilities of XR devices into account, and/or result feedback. This tutorial on Multimedia Retrieval in XR discusses and demonstrates existing solutions and highlights key challenges in this evolving field.

CLIP-based text-to-image retrieval has proven to be very effective at the interactive video retrieval competition Video Browser Showdown 2022, where all three top-scoring teams had implemented a variant of a CLIP model in their system. Since the performance of these three systems was quite close, this post-evaluation was designed to get better insights on the differences of the systems and compare the CLIP-based text-query retrieval engines by introducing slight modifications to the original competition settings. An extended analysis of the overall results and the retrieval performance of all systems' functionalities shows that a strong text retrieval model certainly helps, but has to be coupled with extensive browsing capabilities and other query-modalities to consistently solve known-item-search tasks in a large scale video database.

Knowledge distillation (KD) is a key technique for transferring knowledge from a large, complex "teacher" model to a smaller, more efficient "student" model. Although initially developed for model compression, it has found applications across various domains due to the benefits of its knowledge transfer mechanism. While Cross Entropy (CE) and Kullback-Leibler (KL) are commonly used in KD, this work investigates the applicability of loss functions based on underexplored information dissimilarity measures, such as Triangular Divergence (TD), Structural Entropic Distance (SED), and Jensen-Shannon Divergence (JS), for both independent and identically distributed (iid) and non-iid data distributions. The primary contributions of this study include an empirical evaluation of these dissimilarity measures within a decentralized learning context, i.e., where independent clients collaborate without a central server coordinating the learning process. Additionally, the paper assesses the performance of clients by comparing pairwise distillation averaging among clients to conventional peer-to-peer pairwise distillation. Results indicate that while dissimilarity measures perform comparably in iid settings, non-iid distributions favor SED and JS, which also demonstrated consistent performance across clients.

In this paper, we introduce the fifth release of VISIONE, an advanced video retrieval system offering diverse search functionalities. The user can search for a target video using textual prompts, drawing objects and colors appearing in the target scenes in a canvas, or images as query examples to search for video keyframes with similar content. Compared to the previous version of our system, which was runner-up at VBS 2023, the forthcoming release, set to participate in VBS 2024, showcases a refined user interface that enhances its usability and updated AI models for more effective video content analysis.

VISIONE is a versatile video retrieval system supporting diverse search functionalities, including free-text, similarity, and temporal searches. Its recent success in securing first place in the 2024 Video Browser Showdown (VBS) highlights its effectiveness. Originally designed for analyzing, indexing, and searching diverse video content, VISIONE can also be adapted to images from lifelog cameras thanks to its reliance on frame-based representations and retrieval mechanisms. In this paper, we present an overview of VISIONE's core characteristics and the adjustments made to accommodate lifelog images. These adjustments primarily focus on enhancing result visualization within the GUI, such as grouping images by date or hour to align with lifelog dataset imagery. It's important to note that while the GUI has been updated, the core search engine and visual content analysis components remain unchanged from the version presented at VBS 2024. Specifically, metadata such as local time, GPS coordinates, and concepts associated with images are not indexed or utilized in the system. Instead, the system relies solely on the visual content of the images, with date and time information extracted from their filenames, which are utilized exclusively within the GUI for visualization purposes. Our objective is to evaluate the system's performance within the Lifelog Search Challenge, emphasizing reliance on visual content analysis without additional metadata.

Cross-entropy loss is crucial in training many deep neural networks. In this context, we show a number of novel and strong correlations among various related divergence functions. In particular, we demonstrate that, in some circumstances, (a) cross-entropy is almost perfectly correlated with the little-known triangular divergence, and (b) cross-entropy is strongly correlated with the Euclidean distance over the logits from which the softmax is derived. The consequences of these observations are as follows. First, triangular divergence may be used as a cheaper alternative to cross-entropy. Second, logits can be used as features in a Euclidean space which is strongly synergistic with the classification process. This justifies the use of Euclidean distance over logits as a measure of similarity, in cases where the network is trained using softmax and cross-entropy. We establish these correlations via empirical observation, supported by a mathematical explanation encompassing a number of strongly related divergence functions.

Dimensionality reduction techniques map values from a high dimensional space to one with a lower dimension. The result is a space which requires less physical memory and has a faster distance calculation. These techniques are widely used where required properties of the reduced-dimension space give an acceptable accuracy with respect to the original space. Many such transforms have been described. They have been classified in two main groups: linear and topological. Linear methods such as Principal Component Analysis (PCA) and Random Projection (RP) define matrix-based transforms into a lower dimension of Euclidean space. Topological methods such as Multidimensional Scaling (MDS) attempt to preserve higher-level aspects such as the nearest-neighbour relation, and some may be applied to non-Euclidean spaces. Here, we introduce nSimplex Zen, a novel topological method of reducing dimensionality. Like MDS, it relies only upon pairwise distances measured in the original space. The use of distances, rather than coordinates, allows the technique to be applied to both Euclidean and other Hilbert spaces, including those governed by Cosine, Jensen-Shannon and Quadratic Form distances. We show that in almost all cases, due to geometric properties of high-dimensional spaces, our new technique gives better properties than others, especially with reduction to very low dimensions.

Similarity search is normally defined to be the task of identifying those objects, from a large collection, that are most similar to a further single object presented as a query. Using polyadic queries, a small set of objects are presented to the system, with the intent of finding those objects most similar to all elements of the query set. A few scenarios have previously demonstrated the usefulness of this notion. For example, we may be searching for images similar to a red balloon over a lake. With a single query, it is impossible to tell if the intent is to search for other images of balloons over lakes, or for other red balloons in any background. If instead we could present a system with a few different images of balloons, all of which are either all red, or all over lakes, the similarity search engine may be able to respond more appropriately. In this paper we demonstrate software which permits the user to provide explicit feedback by selecting the best few results from an intermediate set which are best suited to their original information need. A polyadic query can be formed from this set, which should give better results with a minimum of user interaction.